Facebook hired hundreds of contractors to review and transcribe audio clips from users who used its Messenger App, according to a new report from Bloomberg.

Contract employees who were hired by the tech giant to review the audio told the publication they were apprehensive of where the audio was recorded, were not provided details of how it was obtained and were tasked only to transcribe it.

Facebook unveiled the feature to transcribe voice clips to text in 2015 and the company maintains only users who opted in to the feature had their audio clips inspected by the third-party contractors.

However, nowhere in its terms of service or support page does it state that humans would be reviewing and transcribing the audio.

“Voice to Text, uses machine learning. The more you use this feature, the more Voice to Text can help you,” the support page states.

Furthermore, if one person in your chat consented to use the feature, any audio in the thread was subject to review by the third party, Facebook’s support page stipulates.

The contractors, who asked to remain anonymous for fear of losing their jobs, said the transcriptions were used to ensure Facebook’s artificial intelligence correctly interpreted the messages, but argued the practice felt unethical because Facebook never disclosed to users that third parties may review their private audio, which frequently contained vulgar content.

After tech companies including Amazon.com Inc and Apple Inc. faced backlash for collecting audio snippets from its consumers’ devices, CEO Mark Zuckerberg’s company claimed his company ended its use of human workers to listen to the audio clips “more than a week ago.”

“Much like Apple and Google, we paused human review of audio more than a week ago,” the company said Tuesday, noting the users who feel their privacy was invaded by its practice opted in to have their voice chats transcribed when they agreed to use Messenger.

Amazon came under fire in April after reports revealed the company used a global team of workers to listen to Alexa audio requests to improve the software. Apple’s Siri and Alphabet Inc.’s Google Assistant also used human review, but say they abandoned the practice following public scrutiny.

The contractors said TaskUs Inc., a California-based outsourcing firm, is one of company’s Facebook hired to review its users’ audio. The employees are reportedly not permitted to reveal who they work for and are directed to call the client by the code name “Prism,” Bloomberg reports.

TaskUs, a company that also works on election preparation and screens political ads, said “Facebook asked TaskUs to pause this work over a week ago, and it did.”

Facebook states in its policy that it collects “content communications and other information you provide” when users “message or communicate with others,” but omits details about human beings screening the content.

The tech giant also denies collecting audio from its users to determine what content it displays in its news feed, maintaining it “only accesses users’ microphone if the user has given our app permission and if they are actively using a specific feature that requires audio (like voice messaging features.) “

Zuckerberg has even argued in Congressional testimony that the allegations the company collects audio is simply a “conspiracy theory.”

“You’re talking about this conspiracy theory that gets passed around that we listen to what’s going on your microphone and use that for ads, ” Zuckerberg told Congress in April 2018. “We don’t do that.”

Yet, in July, the Federal Trade Commission slapped Facebook with the largest fine the social media network has faced yet, approving a record $5 billion settlement for breaching its privacy policies.  

The FTC found Facebook accessed the data of 87 million users without authorization and shared it with third parties, violating its terms of agreement to notify users when their data is shared.